Accuracy and cost-effectiveness of [18F]-2-fluoro-deoxy-D-glucose-positron emission tomography scan in potentially resectable non-small cell lung cancer.

STUDY OBJECTIVES This retrospective study of patients who were referred for surgical resection of non-small cell lung cancer (NSCLC) assessed the accuracy and cost-effectiveness of positron emission tomography (PET) with radiolabeled [18F]-2-fluoro-deoxy-D-glucose (FDG) in staging mediastinal lymph nodes (MLNs). DESIGN From January 2001 to September 2002, 90 patients with suspected or proven NSCLC who had been referred for curative resection were retrospectively reviewed. All patients were without evidence of metastatic disease. Sixty-nine of the 90 patients had undergone thoracic FDG-PET imaging as part of their evaluation and are the focus of this study. Sensitivity, specificity, accuracy, and positive and negative predictive values for metastasis to the MLN were calculated for CT scanning vs FDG-PET scanning. Four algorithms for staging MLN with mediastinoscopy and/or FDG-PET scan were compared. MEASUREMENTS AND RESULTS Sixty-nine patients underwent preoperative CT and FDG-PET scans, and 32 of 69 patients underwent mediastinoscopy. Fifty-seven patients underwent thoracotomy with complete mediastinal lymphadenectomy. Sensitivity, specificity, accuracy, and positive and negative predictive values for CT scans and FDG-PET scans were 46%, 86%, 78%, 43%, and 87%, and 62%, 98%, 91%, 89% and 92%, respectively. Mediastinoscopy was accurate in 32 of 32 patients (100%). Routine mediastinoscopy remains the most economically reasonable strategy with excellent sensitivity. Selective FDG-PET imaging improved the sensitivity of noninvasive staging for patients with normal MLNs on CT scans. CONCLUSIONS Selective use of FDG-PET imaging improves staging accuracy compared to CT scanning alone and makes it a cost-effective adjunct to the preoperative staging of NSCLC. However, in patients with adenocarcinoma and MLNs of < 1 cm, FDG-PET scanning cannot yet replace mediastinoscopy.